System for detecting resin canals in frankincense trees using multisensor imaging and deep learning

Abstract

A system (100) for detecting resin canal regions in a tree, the system (100) comprising: an image acquisition unit (102) configured to acquire one or more images of a bark surface of the tree using at least one image sensor and one or more additional sensors; a preprocessing unit (104) coupled to the image acquisition unit (102) and configured to enhance one or more images; a feature extraction unit (106) coupled to the preprocessing unit (104) and configured to extract morphological, color, and texture features from one or more enhanced images; a detection unit (108) coupled to the feature extraction unit (106), the detection unit (108) being configured to identify areas corresponding to resin spots, resin drops, cracks, or resin canal structures on the bark surface;and an output unit (110) coupled to the detection unit (108) and configured to produce annotated images and detection results indicating the positions of resin canal areas and the associated detection confidence values, wherein the system (100) enables the automatic identification of resin canal areas in the bark of the tree to support optimized resin harvesting.